Slopwatch: Reckless FUD and Machine-Generated Spam from LinuxSecurity.com, cybersecuritynews.com, and gbhackers.com (Google Boosts LLM Slop About "Linux")
Days ago: Thinking About Abandoning 'Google News' Altogether Due to Easy Poisoning by LLM Slop
Google and so-called 'Google News' continue to yield anti-Linux misinformation. The problem is, and it's a growing source of trouble in fact, a lot of this misinformation was generated by machines, not humans. That means that it's hard to hold someone accountable or say that some particular author sucks. So instead we must focus on whole domains and their reputation. What if LinuxSecurity.com, cybersecuritynews.com, and gbhackers.com are slopfarms and 100% of their output is just garbage that came from LLMs, perhaps curated by and for Microsoft? What would be the impact or implications for so-called 'articles' about "Linux"? Remember how Microsoft feels about Linux. To use a comparison or a political analogy, Russia does not want peace in Ukraine, it wants Ukraine. Likewise, Microsoft does not love Linux, it's busy trying to find ways to "own" it.
Consider what Microsoft did to Linux with Rust. A new article in The Register, published this week (listed here with comment), reminds us of the role Microsoft staff plays inside Linux. These people cause trouble. More such articles got aggregated right here just less than an hour ago.
Now, let's look at new LLM slop about Linux and Rust. cybersecuritynews.com and gbhackers.com seem to be working in pairs again, spewing out almost the same FUD, generated by LLMs with slop for images:
Of course, as usual from these domains, it is LLM slop.
It tells the same lie/s in slightly different ways and 'Google News' has picked both the above (classified as "news"), e.g. when searching for "Linux" just earlier today.
And it doesn't quite end there. There's also this new 'article' about "Cyber threats and Linux vulnerabilities" in a site called LinuxSecurity.com:
Of course it's fake:
There have also many articles about Mozilla Thunderbird 136 lately (partial list), perhaps a "training set" to plagiarise about one week after the real articles came out:
Of course all of this text is LLM slop:
People may be getting fed not only FUD; they're reading garbage produced by programs with no actual comprehension of the words and meanings. █